hash patent by AltNet; Python is prior art?

Tim Churches tchur at optushome.com.au
Sat Jan 15 23:36:28 CET 2005

GerritM wrote:
> ZDnet features an article about the had patent at AltNet
> http://news.zdnet.com/2100-9588_22-5534087.html . Apparantly this issue
> plays already for some time, now the p2p companies are threatened, becuase
> they use hashing to recognize files.

I find it incredibly annoying when journalists reporting on patent 
issues don't bother to quote the patent numbers (or patent application 
numbers) concerned (failure to do so probably means they haven't 
bothered to examine the patent or patent application themselves). In 
this case, a search is made doubly difficult because the patent was 
purchased from someone else, so a name serach can't be done. A quick 
keyworfd search on the USPTO web site revealed some related patents 
issued to Apple and Sony, but hard to identify this particular patent.

> As far as I know hasing is a very old technique used to quickly look up
> information. The use of hashing always requires an idnetity check, because
> the lookup is not unique (mapping a very big amount of possibilities on a
> limited amount of entries). This is a fast and robust way of finding
> information, if the right hashing function is used.
> How can this type of fundamental knowledge be patented? I am afraid this is
> again an example of a total failure of the current patent system.
> Unfortunately these ridiculous examples endanger smaller companies, open
> software activities amd innovation in general.

What you have to remember is that the US (and some countries which have 
been stupid enough to follow its lead, like Japan, Australia and now it 
seems India) permits patents not just on software algorithms and 
techniques, but also on business methods. Thus the "novel" application 
of a well-known and widely-used technique such as hashing to a 
particular field of endeavour (like file sharing on the Internet) can be 
patented (in the US). To be novel, there must be no published prior art 
- but the prior art needs to be specific to the claims of the patent. So 
if the patent says "use of hashing to uniquely identify files in a 
peer-to-peer network via IP over the Internet", then the prior art needs 
to describe exactly that. The Australian government recently passed 
legislation which tightened this test of novelty (thank goodness) by 
allowing separate bits of prior art which describe aspects of a patent 
claim to be combined to some degree when opposing a patent - but the 
test for novelty remains remarkably lax in the US, I beleive.

And I agree 100% with Alex Martelli - Europe and other countries must 
reject software, algorithmic and business method patents and thus create 
a powerhouse of innovation. let the US and fellow-travellers stew in 
their own juice - they will become powerhouses of litigation instead. 
Provided other countries don't recognise software and business method 
patents, the litigation will be confined within US borders, where 
resources can be productivelt spent making television dramas about 
attractive young patent attorneys and plodding, tram-riding patent 
clerks who are really brilliant physicists if only someone would 
recognise their potential. So yes, please write to your MP and MEP and 
protest against the prospect of software and business method patents in 
Europe. Hopefully one day within my lifetime we'll have a governemt here 
in Australia which will roll back the damage done to our patent system 
by trying to make just like the US system, just so we can conclude an 
unbelieveably inequitable free trade agreement with the US. But that's 
our problem.

Tim C

More information about the Python-list mailing list